Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Ocean Optics USB4000 Spectrometer

433 views
Skip to first unread message

MAH2681

unread,
Jul 13, 2007, 1:40:05 PM7/13/07
to
Hello all,
 
I am trying to get an Ocean Optics USB4000 Spectrometer working under Labview.  Several months back I was able to get the USB2000 model working correctly using a set of vi's I came across on the web, however I seem to have found a snag with the newer model.  The vi's I currently use have the option to specify the USB4000 as the model to be used.  There is one vi in particular that calls the DLL.  I tried exchanging the previous function for the USB2000 with the same one that corresponds to the new model, but have no luck.  This causes and error to appear to appear or Labview crashes. 
 
I was hoping maybe someone out there has either used Labview and the USB4000 or could help me out with the problem I seem to be having with my older set.  The spectrometer works fine under the standard Ocean Optics packages.  I could provide vi's if needed for further explanation.
 
Thanks.... 

TWR21

unread,
Feb 7, 2008, 3:40:09 PM2/7/08
to
I recently had similar problems with my USB2000+ from OceanOptics.  I was receiving a "invalid access to memory" error related their dlls. 
The following fix they suggested worked...but don't ask me how :smileywink:
Goto Start -> Run -> Type in "ooidrv.ini" (without the quotes).  This will bring up the .ini file for your oceanoptics device.  Make sure that "Initialized = 1".
Mine was set to 0 and changing it to one miraculously fixed all the .dll errors I was receiving!

Benlyons

unread,
Mar 4, 2008, 2:10:06 PM3/4/08
to
I was having the same problem with the USB4000 vi. I kept getting a message saying "Error 1097 at Call Library Function node... An exception occurred..." and got Ocean Optics gave me the following advice: "Make sure the user doesn?t have a OOIDrv.ini file in C:\Windows.  If he does,
delete it.  If the user uses the USB4000_sample.vi, all the user has to do is
click ?Cancel? and it should work fine.  However, also make sure that the user
has OOIDrv32.dll, NatUSB.dll and JoeDriver.jar all in the USB4000_specific
folder (by default, these three files are in the C:\Program Files\Ocean
Optics\OOILVD\ folder)."A few other things give that Error 1097, for example: reopening and running a LabVIEW project containing the USB4000 drivers a second time without completely shutting down LabVIEW, or if the spectrometer is already connected before turning your computer on.

Shib

unread,
May 26, 2008, 3:10:05 PM5/26/08
to
hi all,cn nybdy help me to get vi for UBS4000 in
Labview6.1 to integrate with DAQ(NI-1407) tghrough USB..........thnx in
advance.........i am using WIn2000(SP4)

CharlyStardust

unread,
May 27, 2008, 3:40:07 PM5/27/08
to
Shib,
You'd better start a thread next time you have a question. This one is out dated.
Labview (and other programming languages) drivers for Ocean Optics spectrometers USB2000 and USB4000 can be purchased at Ocean Optics website. They're not good quality though and may not work with Labview 6.1.
Charly

A_J

unread,
Jul 28, 2008, 2:40:07 AM7/28/08
to
Hi Shib,
I have the generic driver and the programming manual with me. If you need then let me know. I can email them to you.
AJ

Beach Comber

unread,
Aug 11, 2008, 1:40:06 PM8/11/08
to
Hello all,
 
When making a call to the Ocean optics OmniDriver.dll with long integration times and or averaging, it seems Labview cannot do anything else until the OmniDriver "Wrapper" returns.  I would like to include some type of feedback to the user such as a progress indicator and or a blinking control.  Is there a way to do this?
 
regards,
Karl

CharlyStardust

unread,
Aug 11, 2008, 2:40:08 PM8/11/08
to
Hey Karl,
Can you be more specific? I experienced some timeout issues with Omnidriver.dll too. Sometimes the OO spectrometer stays frozen for no reason at initialization.
 
I noticed that to get a corrected signal the spectrometer has to 'run several integration times'. I mean that there is a signal offset that is corrected run after run. I figured out that a bit less than 20 runs are necessary to to get a corrected signal.
I developed a progress indicator with LV whose total time is 20 runs * integration time. It doesn't communicate with the dll it just assess the necessary time to get accurate measurements.
Charly
0 new messages